Original dataset Basset is trained on

[SBNoor]

I am trying to recreate Basset dataset. I want to make sure that is prepare_compendium.ipynb the notebook that recreates that dataset? And if so then when I run the following code:

!cd ../data; preprocess_features.py -y -m 200 -s 600 -o er -c genomes/human.hg19.genome sample_beds.txt

This gives me empty er files. Can you explain why that might be? I am simply trying to recreate the original Basset dataset. Insights will be appreciated.

[davek44]

Hi, I'm better able to support the Basset method here in my new DNA conv net repository. Try this path to recreate the dataset https://github.com/calico/basenji/tree/master/manuscripts/basset